Evaluating a 3D Face Reconstruction Model Using 2D Images
Author Information
Author(s): Xiao Yujia, Mao Bochun, Nie Jianglong, Liu Jiayi, Wang Shuo, Liu Dawei, Zhou Yanheng
Primary Institution: Peking University School and Hospital of Stomatology
Hypothesis
The null hypothesis was that there were no statistically significant differences between the face scan models and 3D reconstruction model in clinical accuracy and feasibility.
Conclusion
The study found that the 3D reconstruction model achieved high accuracy with a mean deviation of approximately 2 mm compared to face scans.
Supporting Evidence
- The RMSE of the reconstruction model was 2.00 ± 0.38 mm.
- High accuracy was achieved for the forehead, nose, upper lip, paranasal region, and right cheek.
- Significant differences were found in the nasolabial, nasal, and nasofrontal angles.
Takeaway
This study shows how we can turn flat pictures of faces into 3D models, which can help doctors understand and explain facial features better.
Methodology
The study used frontal and lateral images to reconstruct 3D facial models and compared them with 3D scans to evaluate accuracy.
Potential Biases
The study may have selection bias due to the exclusion criteria.
Limitations
The study had a small sample size and excluded patients with significant facial asymmetry or scars, limiting its applicability.
Participant Demographics
The participants were 14 females and 9 males, average age 20.70 ± 5.36 years, all Han Chinese.
Statistical Information
P-Value
p<0.05
Confidence Interval
95% CI: 1.84–2.17 mm
Statistical Significance
p<0.05
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website